Semantic Modeling

1
Semantic Modeling Data ManagementMDM Project
Outcome5/04/05
2
Objectives

• Explain MDM Project approach
• Why build a Semantic Model before Logical Model
• How to exploit the Semantic Model
• Learn from the experience
• Difficulties encountered
• Best practices
• Draw up the project outcome
• Benefits of Semantic Modelling
• Benefits for the IS

3
Contents

• The Approach
• The theoretical basis
• Process
• Semantic Modelling
• Principles objectives
• Illustration
• Benefits
• Semantic Model Transformation
• Logical Data Model
• Service Oriented Architecture
• The Plan
• The story of the Project
• Continuing on post-MDM project

4
Agenda
5
Contents of Part 1
1
The Approach

• "Aspects" of the Enterprise System
• Theoretical basis to the approach
• Principles objectives of Semantic Modelling

6
The Enterprise System
The Approach
Knowledge the core business, notions
How to connect them? How to optimise them?
Activitybusiness processes
Software Tools
Equipment
7
"Aspects" of the Enterprise System
The Approach
The Enterprise System Topology
8
The Service transformation chain
The Approach
Definition et description
Design
Origin
Design Conditions
Deployment et execution
9
The MDM context
The Approach

• A multiple aspect approach
• The MDM project respects the requirements
• Provide signification to the data
• Convergence on a data perspective
• To do this, the approach is based upon the
  following principles
• Abstraction
• Conceptual view
• Signification
• Define the modelling elements as rigourously as
  possible
• Structure
• Improve the quality of the system encapsulation,
  limited coupling
• Prepare for a Service Oriented Approach

10
Products
Approach

• The final deliverable is the Logical Data Model
• Scope needs of the CORSO project
• LDM Relational Model
• An intermediate deliverable the Semantic Model
• Other derived products
• To facilitate the use of the models

LDM
SEM
Hypertext
Dictionary
11
The Process
The Approach

• "Innovation" Approach

Inception
Exploration
Crystallisation
Specification
Production
Communication
Consolidation
Collection
Collection
...
1
2
3
12
The Approach
Data Management project
Scope
Semantic Model
Depth
Logical Data Model
CORSO project
Technical architecture Software
13
The Approach
Data Management project
Scope
Production
phase
Exploration phase
Crystallisation
Semantic Model
Depth
Logical Data Model
Communication phase
CORSO project
Technical architecture Software
14
Exploration phase
The Approach
Exploration PhaseCelesio Selecting the relevant
information from the legacy
Exploration PhaseUnilog Modelling the
semanticsfrom use-cases description
Logical data description
Draft Sem Mdl
Production PhaseMatching the Sem Mdland the
existing data
Comm. PhaseDeriving the LD Mdl from the Sem Mdl
Enhanced Sem Mdl
LD Mdl
15
Semantic Model and Use Case View
The Approach

• One of the representation inside the
   Pragmatical  aspect
• Functional approach
• Local point of view

Organisation View
Use Case View
16
Method deliverables
The Approach

• Users Guide to Semantic Model
• MDS03-UserGuide.doc
• Templates for the collected documentation

17
Contents of Part 2
2
Semantic Modelling

• Principles and objectives of semantic modelling
• Difference between classical and semantic
  modelling
• Illustration
• Genericity
• Transfer order

18
Principles objectives
Semantic Modelling

• Format
• UML language class diagram, state diagram
  collaboration diagram
• Data Dictionary
• Synonyms, keywords, definitions and comments
• High level of abstraction
• Improves expressiveness
• Provides a basis for modelling for follow-on
  projects
• Data oriented (with an object-oriented approach)
• Reminder It is not be a complete model (few
  operations and constraints will be captured)

19
From classical data model
Semantic Modelling
20
towards semantic model
Semantic Modelling
21
Genericity
Semantic Modelling

• The generic concept  Offering 
• Allow to order services as well as product by the
  same process
• Avoid structural redundancy
• Factorizing the common properties
• The generic concept of actor
• Customer, Supplier, Entity, Person
• A given actor can act in many ways
• Both as customer and supplier
• As a worker
• A supplier or a DC can order products

22
Genericity (continued)
Semantic Modelling

• Transfer order

issuer
23
Illustration of transfer order
Semantic Modelling
issuer
24
Pricing the only way to record a price
Semantic Modelling
25
How to express pricing conditions
Semantic Modelling
26
Pricing detailed solution
Semantic Modelling
27
Benefits
Semantic Modelling

• Goal of semantic modelling
• Capture the very meaning of the business
• Help the convergence
• Characteristics of semantic models
• Expressiveness
• Genericity
• Adaptability
• Object-oriented approach versus functional
  approach
• The case of pricing

28
Highlights on the approach
Semantic Modelling

• Object-oriented philosophy
• Versus data-oriented approach
• A condition for restoring the meaning
• The model shows some operations as sample
• The state machine expresses the dynamic
• The quest for genericity
• Provide the more of services with the less of
  terms
• Separation of concerns
• The abstraction levels
• Get rid off the technical concerns (for a while)
• Focus on the essential core
• Modelling rules

29
Modelling tools
Semantic Modelling

• Standard notation
• UML (Unified Modelling Language)
• The diagrams we used
• Class diagram
• Object diagram
• State diagram
• Pre-modelling
• The data dictionary
• The traceability
• Reference to the collected documents

30
Contents of Part 3
3
Semantic Model Transformation

• The logical aspect and the service oriented
  architecture
• Principles and methods
• Services and data
• Layers
• Transformation
• The by-products of the semantic model
• The Logical Data Model
• Other possible transformations

31
Service Oriented Architectureprinciples
Semantic Model Transformation

• The Service layer masks the Data layer
• The Services are ordered in the logical
  architecture
• Conception of Services follows on from
  Urbanisation decisions
• Services and data derive from model of the upper
  abstract level
• The transformation from an aspect to an other

32
Encapsulating
Semantic Model Transformation
Service
logical blocks
Data layer
The service layer hides the data layer
33
Structuring
Semantic Model Transformation

• The Services are ordered in the logical
  architecture
• Several levels of composition

Fabrique logique
Atelier logique
Machine logique
Services
34
Layering the System
Semantic Model Transformation

• At a logical level
• Not to be confused with the technical
  architecture model

35
Origin of Logical Services
Semantic Model Transformation
Présentation
Machines logiques Organisation
Machines logiques Métier
36
Transformation Chain (1)
Semantic Model Transformation
Technical Aspect
Logical Aspect

• A1 Java, Oracle
• A2 RPG, DB2
• Etc.
• development rules

 Machine  M
Visible structure
 Service  sv
Software Aspect
DB1
DB2
37
Transformation Chain (2)
Semantic Model Transformation
Hardware Aspect
Software Aspect
DB1
Physical Nodes
DB2
Physical Aspect
DB1
DB1
DB2
38
Logical Architecture Graph
Semantic Model Transformation
39
Logical Architecture Graph (detailed)
Semantic Model Transformation
40
Subjects for the Logical Model
Semantic Model Transformation

• Logical / Technical discussions
• Examples
• Transaction management
• Error management
• Event management
• Regarding the modelling of the Logical Model
• Object identification
• Keeping history traces
• Data reconciling

41
Object identification (1/3)
Semantic Model Transformation

• The problem
• A unique system for all countries
• A unique system for the classes inside the same
  inheritance tree

42
Object identification (2/3)
Semantic Model Transformation

• An example

43
Object identification (3/3)
Semantic Model Transformation

• The solution
• A universal system for object identification
• Valid for all countries
• This is a question for the logical design
• Unless it receives a technical answer

44
Keeping history traces (1/2)
Semantic Model Transformation

• On many concepts
• Creation date, update date and so on
• User responsible for the action

45
Keeping history traces (2/2)
Semantic Model Transformation

• It is not relevant in the semantic model
• Risk of overloading the core model
• Another issue for the logical design

46
Data reconciling (1/2)
Semantic Model Transformation

• The problem
• Existing data in different structures
• Must converge to the future and unified system
• Either feeding the future system
• Or providing it with information as requested

47
Data reconciling (2/2)
Semantic Model Transformation

• Two options
• Conversion
• Transforming data, migrating to the new database
• One-shot, on a project-time basis
• Reconciliation
• Thanks to EAI possibilities
• On a run-time basis
• This is part of the logical/technical discussion
• It can be decided by parts
• It must be set in the urbanisation roadmap

48
Contents of Part 4
4
The Plan

• Milestones
• Collected documents
• Continuing on post-MDM project
• Consolidation
• Possible works to be done

49
Milestones
The Plan
50
Collected documents
The Plan

• Collected documentation towards Semantic Model

• The accuracy of the collected documents has been
  poor
• See metric.xls file for details

51
Convergence Metrics
The Plan

• The Semantic Model attributes

UK
18
FR
114
20
12
69
29
DE
64
312
52
Continuing on post-MDM project
The Plan

• Explaining and maintaining the semantic model
• The MDM project ends with the delivery of the
  logical data model
• It can carry on with the  Consolidation phase 
• Change process on SEM and LDM

Inception
Exploration
Crystallisation
Specification
Production
Communication
Consolidation
53
Possible works
The Plan

• On the semantic model
• Complete the State machines
• States Transitions
• Identification of all the Operations
• Before deriving the services
• Complete improve upon comments
• More explanations
• From the use case description
• Bridge the Use Case View and the SEM
• By means of connections to the relevant modelling
  elements
• By means of new diagrams
• Structure the Use Case View
• By eliminating redundancy

54
Conclusion

• Outcome of the MDM project
• Benefits of Semantic Modelling
• Convergence
• Simplification
• A start for the business repository
• Benefits for the IS
• A source for deriving the services in SOA
• Planning, means and budget
• Kept on track
• Absolute need for a UML tool
• Difficulties
• Tools
• Cultural change
• State of the input documentation
• Scope